⬅️ Project: Conquest Without Europeans · Cases: Africa · Asia · Argument: Opposition Case · Standing rules: Citation Audit
Which numbers in this project are safe to say out loud, which need the caveat spoken with them, and which will lose you the room. Plus an honest account of what has and has not been verified against a primary source.
Verification status — updated after a library sweep
Five sources have now been opened and read against your own NAS, vault, and Zotero: Perdue (China Marches West), Millward (Eurasian Crossroads), Frahm (Assyria), Karsh (Islamic Imperialism), and Austen (1992). Claims sourced to those five are marked ✅ verified in the case files and are cleared for use.
Everything else remains unverified — written from general historical knowledge, not yet opened. Per the standing sourcing rules (Citation Audit), those claims are not cleared for the Substack or a live debate.The sweep changed three things, and one of them was a claim I told you to memorize. See ⬇️ Corrections from the verification pass. This is the workflow doing its job: the number I gave you for the Zunghars was wrong, and the source that corrects it was already on your shelf.
Corrections from the verification pass
| Claim as first written | What the source actually says | Status |
|---|---|---|
| ”Perdue estimates roughly 80% of ~600,000 Zunghars died” | The figures are Wei Yuan’s (Shengwuji), quoted by Perdue at p. 285 — not Perdue’s own estimate. Breakdown: 40% smallpox · 30% killed by the army · 20% fled · 10% enslaved. So ~70% dead, 90% removed. | ❌→✅ corrected in the Asia note |
| “Perdue treats it as genocide” (stated as settled) | True — Perdue, p. 284: “they had never attempted ethnic genocide before.” But James Millward, whose book you also own, calls that framing “an exaggeration” (Eurasian Crossroads, p. 94) and cites Levey (2014) for there being no Zunghar genocide. | ⚠️ now a flagged dispute, not a settled fact |
| ”~7 million trans-Saharan (Ralph Austen)“ | Austen’s 1992 paper — the one you own — says “I have deliberately avoided drawing up grand totals” and states his revision will lower the Mediterranean totals. The round figures come from his superseded 1979/1988 censuses. | ⚠️ route totals pulled from the Africa note |
| “Oded: 157 deportations, ~4.5 million” | Not checked. Frahm (in your vault) gives ~600,000 deportees under Tiglath-pileser III alone, across ~50 operations, with his own caveat that it “may be somewhat exaggerated, but probably not by that much.” | ✅ Frahm substituted; Oded left flagged |
The lesson worth keeping: in all four cases the correction made the argument better. Conceding that Millward disputes Perdue, and that Austen refuses to total his own numbers, is more persuasive than the confident version — because it shows you read past the citation.
Landmines already sitting in your vault
Found during the sweep. These are notes you already have that would lose you an exchange if quoted.
Islam Invasion of India — do not quote any figure in this note
The note opens by attributing to the Mughal-era historian Firishta the claim that “over 400 million Hindus got slaughtered”, and states that “India’s population is said to have been around 600 million at the time of Muslim invasion.”
This is impossible on its face. The entire world population around 1000 CE is estimated at roughly 250–300 million. India’s was on the order of 75–110 million. A claim that India alone held 600 million people, and lost 400 million of them, is off by a factor that any opponent with a phone can check in fifteen seconds — and it will be used to discredit everything else you say.
The rest of the note is a copy-pasted polemical essay in the same register (“contaminated by the association with Islamic ‘culture’”, “savaged”). It is not a source; it is the kind of thing the ❌ list exists to stop.
What to use instead: the Asia note's India section — Eaton’s ~80 documented temple desecrations, and the sultans’ own chroniclers boasting of massacre and enslavement. Smaller, true, and unanswerable.
Notes whose citations are all Wikipedia
Non-European Colonialism (already flagged and superseded) · Arab Slave Trade (five Wikipedia links, nothing else) · Zanzibar Revolution · Islam Caliphates, Conquering, Colonizing · Mongol Empire · African wars · 1698-1730s - Oyo invasions of Dahomey · 1076 - Almoravid conquest of Ghana Empire
These are fine as finding aids — they tell you what happened and what to look up. They are not citable, and per your own standing rule they shouldn’t be the basis of a public claim. The Oyo–Dahomey and Almoravid–Ghana notes matter most, because both are cited in this project’s case files.
How the rest of this note works
It separates the claims that need no source from the ones where a single unverified number can lose an exchange. Work down the ❌ and ⚠️ lists first; the ✅ list can wait.
❌ Do not use — these will lose you the argument
Each of these circulates widely in exactly the kind of material that would supply this argument. Each collapses under a well-read opponent. Knowing why they’re wrong is worth more than the numbers ever were.
| Claim | Why it fails | Use instead |
|---|---|---|
| ”80 million Hindus killed by Muslim rule in India” | Traces to K.S. Lal, Growth of Muslim Population in Medieval India. Rejected across the professional field — the method back-projects population figures with no defensible basis. It is a marker of unserious sourcing and a specialist will recognize it instantly. | Richard Eaton’s ~80 documented temple desecrations, and the sultans’ own chroniclers boasting of massacre and enslavement. A small verified number beats a large disputed one. |
| ”Timur killed 17 million” | No sound derivation; roughly 5% of the estimated world population of the time, asserted without a basis anyone can trace. | The specific documented massacres from his own court chronicles — Isfahan, Delhi, Baghdad. Far more damning and unattackable. |
| ”The An Lushan Rebellion killed 36 million” | An artifact of comparing Tang census figures (≈53m in 754, ≈17m in 764). The census measured households the state could tax, and the state had collapsed. Deaths, flight, and administrative failure are conflated. | ”One of the deadliest conflicts in history; the Tang tax census fell by roughly two-thirds, which reflects state collapse as well as death." |
| "The Mfecane killed 1–2 million” | See below. The figure derives from settler-era accounts with a direct interest in depicting a self-emptied land. | Don’t use a figure. Use Zulu state formation, the Ndebele subjugation of the Shona, and say the tolls are contested. |
| ”Mongol conquests killed 40 million” | Back-derived from Chinese census data with the same problem as An Lushan, then rounded upward through repetition. | ”By most estimates the deadliest imperial project before the twentieth century,” plus the documented city destructions. |
| ”60,000 (or 40,000) temples destroyed in India” | A polemical figure with no scholarly basis. | Eaton, again. |
| ”Mansa Musa’s spending crashed the Egyptian economy for twelve years” | Al-‘Umari does report that the price of gold fell in Cairo after the 1324 hajj. The “twelve years” and the dollar-value versions are internet accretion. | ”Contemporary Arab accounts report the gold price in Cairo fell after his visit.” |
⚠️ Real, but state the caveat in the same breath
Using these without the caveat is how a true claim gets treated as a false one.
| Claim | The caveat that must be spoken |
|---|---|
| Arab/Islamic slave trades moved 11–17 million people | ⚠️ Do not attribute route totals to Austen. His 1992 paper (which you own) refuses grand totals outright and says his revision lowers the Mediterranean figures; the round numbers come from his superseded 1979/1988 censuses. Argue duration and abolition dates instead — those are unattackable. High-end polemical totals (N’Diaye and others) are contested. |
| Nanjing Massacre deaths | Genuinely contested: roughly 40,000 to the official Chinese 300,000. The event is not in dispute outside Japanese revisionist circles. Give the range. |
| Great Leap Forward: 45 million | Dikötter’s figure. Yang Jisheng gives ~36m; other serious estimates run to 15–30m. Say “estimates from 15 to 45 million, with Dikötter at the top of the range.” |
| Zunghar deaths | ✅ Verified. Not “80% died” — Wei Yuan via Perdue p. 285: 40% smallpox, 30% killed by the army, 20% fled, 10% enslaved = ~70% dead of ~600,000. Say it that way; the breakdown is more convincing than the round number. And flag Millward’s dissent before your opponent does. |
| Yangzhou: 800,000 dead | Wang Xiuchu’s eyewitness figure, almost certainly inflated as such figures usually are. The massacre is not in doubt. Say “the eyewitness account claims 800,000; the figure is likely inflated.” |
| Timur executed 100,000 captives at Delhi | From his own side’s chronicles, which boast. Attribute it: “his court chronicle records.” |
| Bai Qi buried 400,000 alive at Changping | Sima Qian, writing ~150 years later. Attribute to the Shiji. |
| Maratha raids killed 400,000 in Bengal | Contemporary estimate from the Maharashtra Purana; not a count. |
| Crimean raids enslaved ~1.75m Slavs, 1468–1694 | Alan Fisher’s estimate; others differ. The system is thoroughly documented; the total is a reconstruction. |
| Three Alls Policy: 2.7 million | Attributed to Mitsuyoshi Himeta; not a consensus figure. |
| Sudan’s population halved under the Mahdiyya | Widely repeated, poorly sourced. The 1889 famine and the campaigns are solid. |
| Bangladesh 1971: 3 million | Ranges from ~300,000 to 3 million and is politically charged in both directions; Sarmila Bose’s low revision is itself heavily criticized. Give the range and note the dispute. |
| The Almoravids conquered Ghana in 1076 | Directly challenged by Conrad & Fisher, “The Conquest That Never Was” (History in Africa, 1982–83). Do not lead with it. The vault’s 1076 - Almoravid conquest of Ghana Empire states the old version. |
| Europeans controlled ~84% of the land surface by 1914 | Commonly attributed to D.K. Fieldhouse; primary source not opened. Say “a commonly given figure” or drop it. |
✅ Safe to assert
These are well-established, uncontroversial in the field, and don’t require a number to do their work.
- Qianlong ordered the destruction of the Zunghars, and the emptied land was resettled and named Xinjiang, “New Frontier.” The archival basis is Perdue’s, from Qing sources; the name is simply a fact about the map.
- Menelik II roughly doubled Ethiopia’s territory in the 1880s–90s and installed the neftenya-gabbar settler system over the conquered south. Contested in interpretation, not in fact.
- The Sokoto Caliphate held one of the largest enslaved populations in the world by 1900. Lovejoy; Lovejoy & Hogendorn.
- Slavery was legal in Saudi Arabia and Yemen until 1962, Oman until 1970, and Mauritania criminalized it only in 2007. Clean, checkable, and it ends the “that was all long ago” line by itself.
- Japan established a Colonization Agency for Hokkaido in 1869 and passed the Hokkaido Former Aborigines Protection Act in 1899; the Ainu were recognized as indigenous only in 2008. Statutory record.
- The Ryukyu Kingdom was annexed and abolished in 1879.
- North African Christianity was extinct as an organized presence by roughly the 12th century; Arabic replaced Latin, Punic, and Coptic. Amazigh gained official status in Morocco in 2011 and Algeria in 2016.
- The Yuan dynasty operated a four-tier legal ethnic hierarchy.
- The mimizuka in Kyoto contains the noses of roughly 38,000 Korean and Chinese dead and is still standing.
- Lê Thánh Tông destroyed the Cham capital in 1471; Champa was finally absorbed in 1832; Minh Mạng ruled Cambodia as Trấn Tây Thành, 1834–41, with forced adoption of Vietnamese customs.
- Oyo made Dahomey a tributary state from the 1730s to 1823.
- The Ottoman devşirme levied Christian children from the Balkans for roughly three centuries.
- The Armenian Genocide killed approximately 1.5 million.
- The Cambodian genocide killed roughly a quarter of Cambodia’s population and was ended by a Vietnamese invasion.
- The Dungan Revolt (1862–77) killed on the order of 8–12 million. The range is wide but the event’s scale is not seriously disputed.
The Mfecane problem
Worth its own section, because it is the trap in this subject.
The old account: Shaka’s military revolution set off a chain of wars and migrations across southern Africa from roughly 1815 to 1840, killing one to two million people and emptying the interior — which is, conveniently, what the Voortrekkers found when they arrived.
Cobbing’s attack. Julian Cobbing, “The Mfecane as Alibi: Thoughts on Dithakong and Mbolompo,” Journal of African History 29 (1988) argued that the concept was an alibi: it presented southern Africa as having depopulated itself through African savagery, thereby legitimizing white settlement of “empty” land, and it obscured the actual drivers — Portuguese slave trading out of Delagoa Bay and Griqua and settler raiding from the Cape.
The counter-attack. Cobbing was substantially criticized in turn — by Elizabeth Eldredge, John Wright, and across Carolyn Hamilton’s edited volume The Mfecane Aftermath (Witwatersrand, 1995) — for overstating the slave-trade evidence and for underrating genuine internal drivers, including drought, population pressure, and the ivory and cattle trades.
Where the field ended up: not back at the old story. The consensus that survives is that Zulu state formation was real and violent, that the causes were multiple and included external commercial and raiding pressure, and that the dramatic death tolls do not have a defensible basis.
The practical rule
Do not use the Mfecane as a headline case. It is the one place in this project where “your sources are colonial” is correct, and an opponent who knows Cobbing will use it to discredit the other thirty cases by association.
If it comes up, get there first: “The Mfecane is actually a bad example, and here’s why — Cobbing showed the death tolls were an alibi for settlement. Which is exactly the standard I’m asking you to apply to Zungharia.” Conceding a weak case unprompted buys enormous credibility for the strong ones.
Census-derived death tolls
A single methodological point that disposes of several bad numbers at once, and is worth being able to explain in twenty seconds.
Pre-modern states counted households liable for tax and corvée, not people. When a census figure collapses after a war, the drop combines: actual deaths · refugees who fled the taxable jurisdiction · households that survived but stopped being registered because the administration collapsed · deliberate evasion once enforcement weakened.
This affects the An Lushan figures, the Mongol totals for China, and several Chinese rebellion estimates. It does not mean nothing happened — these were catastrophes on any reading. It means the specific headline numbers are not measurements, and asserting them as measurements is how you lose an argument you were winning.
The twenty-second version: “Those figures come from tax registers, not death certificates. The Tang state lost two-thirds of its registered households — some died, some fled, and some just stopped being counted when the government fell apart. It was a catastrophe; it wasn’t a census of the dead.”
Verification checklist
Nothing below has been opened. Ordered by how much a bad citation would cost.
Priority 1 — the load-bearing claims
| Source | Verifying what | In vault? |
|---|---|---|
| Peter C. Perdue, China Marches West (Harvard, 2005) | Qianlong’s edicts; the Wei Yuan figures; Perdue’s use of “genocide” — pp. 282–87 | ✅ on your NAS, read, verified |
| James Millward, Eurasian Crossroads | The counter-case against Perdue’s framing — p. 94, and the Levey 2014 citation | ✅ on your NAS, read, verified |
| Efraim Karsh, Islamic Imperialism (Yale, 2013) | The whole Arab-conquest-as-imperialism argument; Introduction pp. 1–10 | ✅ on your NAS, read — now the project’s best framing source |
| Eckart Frahm, Assyria | Tiglath-pileser III’s ~600,000 deportees; the Arrapha governor’s letter | ✅ in your vault, read, verified |
| Paul Lovejoy, Transformations in Slavery (Cambridge) and Lovejoy & Hogendorn, Slow Death for Slavery (1993) | Sokoto’s enslaved population size and its ranking | ❌ acquire |
| Richard M. Eaton, “Temple Desecration and Indo-Muslim States” (2000) | The ~80 documented desecrations and his argument about political targeting | ❌ acquire — short, freely available, do this one first |
| Julian Cobbing, “The Mfecane as Alibi” (JAH 29, 1988) | The alibi thesis, before you concede it in public | ❌ acquire |
| Ralph Austen, 1992 | The route totals — turns out he declines to give them | ✅ in Zotero, read — and it corrected the note |
| Ehud Toledano, The Ottoman Slave Trade and Its Suppression | Who ended the Ottoman trade, when, and against what resistance | ⭕ on your NAS, not yet opened — highest-value unread item |
| Matthew Hopper, “Slaves of One Master” | The 19th-c. Indian Ocean and Gulf trade | ⭕ in Zotero, not yet opened |
Priority 2 — the interpretive claims
| Source | Verifying what |
|---|---|
| Bustenay Oded, Mass Deportations and Deportees in the Neo-Assyrian Empire (1979) | The 157 deportations and the ~4.5 million figure |
| Asafa Jalata, Oromia and Ethiopia (1993); Holcomb & Ibssa, The Invention of Ethiopia (1990) | That the “Abyssinian colonialism” framing is Oromo scholarship, stated in those terms |
| Conrad & Fisher, “The Conquest That Never Was” (History in Africa 9 & 10, 1982–83) | The Ghana/Almoravid revision |
| Mahmood Mamdani, When Victims Become Killers (2001) | Pre-colonial vs. colonial racialization in Rwanda |
| Audrey Truschke, Aurangzeb (2017) | The Hindu-administration and temple-patronage claims — needed for the one-standard argument |
| Bernard Lewis, Race and Slavery in the Middle East (1990) | The Zanj material used against the racialization objection |
| Frederick Cooper, Colonialism in Question (2005) | His argument against concept-stretching — already in the vault ✅ → hub library |
| Patrick Wolfe, “Settler Colonialism and the Elimination of the Native” | The exact wording of the “structure, not an event” and “logic of elimination” formulations — already in the vault ✅ → Settler colonialism and the elimination of the native.pdf (file not included in this release) |
Priority 3 — primary sources worth reading rather than citing secondhand
Ibn Khaldun, Muqaddimah (Rosenthal translation) on the Banu Hilal · the Ta’rikh al-Sudan (Hunwick’s translation, Timbuktu and the Songhay Empire) · Piye’s Victory Stele (translated in Lichtheim, Ancient Egyptian Literature vol. III) · Ibn Battuta on Mali · al-‘Umari on Mansa Musa · Wang Xiuchu, “Ten Days at Yangzhou.”
Where to look first
The NAS library and the Zotero corpus. Eaton’s article and Cobbing’s are journal pieces and the fastest wins on this list — both are short, both are load-bearing, and both are the kind of thing an opponent has read.
Also see
Conquest Without Europeans · Conquest Without Europeans - The Opposition Case · Citation Audit · Genocide percentages · Continents by Death Toll · Debate/Genocide/ (the definitional work)